how is a 2.0ghz amd better than say a 3.8ghz intel cpu?

I know this question may seem kinda noobish but I really dont understand how for example an AMD 3200+ running at say overclocked to 2.5 is better than say an intel (i dont now the processor name) overclocked to 3.8? I heard that amds are more efficient but like what is the down dirty truth of it all? I just want this info so i can explain to my friends why my 3200+ is running only 2.5 and their intel is running like super high 3.8

First of all, the cores of AMD and Intel processors are different. Because of this, you cannot compare the speeds of the two directly. In general, all AMD processors do more work per clock cycle than Intel processors. So that is why 2.2 Ghz on an AMD processor might compare to 3.0 Ghz on an Intel P4.

Also, keep in mind that the Athlon 64 possesses an on-die memory controller. Basically, standard systems are designed so that the memory interacts with the Northbridge, which takes time. We enthusiasts have attempted to expidite this process through tweaks such as latency reductions, but the fact of the matter is that, no matter how low the timings go, a small window will always exist when the memory and Northbridge interact (like on an Intel system for example).

The Athlon 64, on the other hand, places the memory controller on the processor itself. This significantly reduces the amount of time it takes for the system to access the memory, and it is also why K8 systems are so fast. Processes that access the memory also benefit from this sort of architecture, which is why Athlon 64's are preferred for gaming at the moment. Furthermore, the on-die memory controller also alleviates many of the weaknesses in AMD systems. Specifically, K8 processors can now run asynchronously without taking signifcant performance penalities, and timings are no longer as pertinent as they once were on earlier systems.

Simply put, people who assume that a 3.xx Ghz Intel is better know very little about computers. As a matter of fact, it plainly corroborates that they are nothing but puppets to the Mhz myth. It is true that clock speeds matter (sometimes), but other features can be just as signifcant, such as cache size or hyperthreading.

Next time your friends make such foolish comments, take pride in the fact that you are generally running a faster system, even though it is technically clocked 1.3 Ghz lower than their own.

deception``

^^^well said, deception

That's...partially true.

AMDs have always been more "efficient" processors than Intels, but in a different way than the AMD fanboy might imagine.

AMD's pipeline is more efficient, meaning that there are less errors along the pipeline which means less packets sent back to the beginning for reprocessing. Intel on the other hand has a very very fast pipeline, but errors are more common. However, the speed on intels makes up for this. With more speed, you effectively void the rule that having errors is bad, as the speed of your pipeline makes up for this.

Eventually, if AMD could get their pipeline up to Intel speeds while maintaining it's efficiency (which may never happen), there is a chance AMD could hold a pretty big lead on intels. Until then, AMD will have to do more of the same and just keep on improving their pipeline.

7

Seven said:

That's...partially true.

AMDs have always been more "efficient" processors than Intels, but in a different way than the AMD fanboy might imagine.

AMD's pipeline is more efficient, meaning that there are less errors along the pipeline which means less packets sent back to the beginning for reprocessing. Intel on the other hand has a very very fast pipeline, but errors are more common. However, the speed on intels makes up for this. With more speed, you effectively void the rule that having errors is bad, as the speed of your pipeline makes up for this.

Eventually, if AMD could get their pipeline up to Intel speeds while maintaining it's efficiency (which may never happen), there is a chance AMD could hold a pretty big lead on intels. Until then, AMD will have to do more of the same and just keep on improving their pipeline.

7

Click to expand...

I think you have things a bit backwards, Seven.

AMD's are actually more efficient because their pipeline is much shorter than the typical Intel processor.

With each and every new P4 (Willy, Northwood, Prescott, etc.) Intel continues to lengthen the pipeline as it seeks to create faster and faster processors. Coincidentally, it takes newer P4's greater increases in clock speeds before their features such as increased cache size are notable. It is for this reason that Prescotts generally do not shine until they reach above 3.6-3.7 Ghz.

Think about it as two highways:

The AMD highway is much shorter than that of the Intel. Because the highway is so short, it takes less time for a car to travel the full distance of the highway; hence, there is less of a need for the highway to be very long.

The Intel highway, on the other hand, is much larger than that of the AMD highway. This can be represented by their clock speeds. However, given that the highway is much longer, it takes a car longer to reach the full length of the highway than that of a comparable AMD. It is for this reason why it requires much higher frequencies for an Intel processor to equal that of an AMD (and vice versa).

I appreciate your thought on the matter, but it helps if you have the basics down pat.

Oh, and I'm not a fanboy, either.

deception``

i think we need a sticky for this topic or something so ppl dont post it so much! its posted all the time.. soeone like deception should make a thread that explains it for all the intel noobs to read.. haha

Amen!!!

deception`` is right

AMD has a shorter pipeline so it can do more work per clock cycle

Intel has a longer pipeline so it does less work per clock cycle but can reach higher speeds.... also Intel can have HT because of the longer pipelines if AMD was to have some sort of HT it wouldnt work aswell because the shorter pipes

As a ocforum member put it once.

Image you are carrying a bucket of water from a well to a fire
AMD can carry a huge bucket but very slowly
Intel carries small buckets but much faster

This is kind of a bad way to put it, it sounds cool.

how i love the forums
so full of enriching info

How detailed/complex of an answer do you want? You can get very, very detailed comparing the two processors, but you've already been given a pretty good intuitive explanation. I'll just add a little bit more for a bit more clarity. I won't bother going to far because your friend won't understand it anyway, but if you want more detail, I'd be happy to provide some information.

Due to Intel's marketing and mass market ignorance(no offense to joe sixpacks out there, I don't think they're stupid, just uninformed and for most, they don't need to be) the perception is that somehow mhz show how much work a cpu is doing, when in reality, it is only part of the formula. The other main part is how many actual instructions are being completed for every "tick" of the processor or for every hertz.

The Athlon64's perform more instructions for every hertz than The P4's(this is due to the pipelining, but I won't get into that, you've already been kinda told about that anyway), so if each were clocked at the same speed, say 2ghz, the A64 would actually be doing more work because it is doing more instructions at the same speed. Add to that the above mentioned onchip memory controller and you end up with a much more efficient cpu. There are still a lot of other things such as branch predicting, cache influences, bandwidth effects, and others you could talk about, but that's beyond what you need.

What's up with the name calling from someone with 7 stars?

Other than that, Awesome information, and a joy to read/learn.

Excellent explinations deception`` and ]-[itman.

comparing clock speeds of intel and amd !:1 is like comparing the gas mileage of a car to a truck, both get the job done but one can do it more efficently

Sorry for this reply that kind of does not pertain to the original post, but im intrigued to learn more about how cpu's actually work...is there a good website that can explain it nicely? But of course for a 16year old it cannot be so hard to understand that i dont comprehend a single word of it
Thanks, Ps2cho

ps2cho said:

Sorry for this reply that kind of does not pertain to the original post, but im intrigued to learn more about how cpu's actually work...is there a good website that can explain it nicely? But of course for a 16year old it cannot be so hard to understand that i dont comprehend a single word of it
Thanks, Ps2cho

Click to expand...

I'll round up some sites for you and pick out the ones to start with, it might take me a day or two(kinda been real busy lately), but I'll pm them to you when I get them.

deception`` said:

I think you have things a bit backwards, Seven.

AMD's are actually more efficient because their pipeline is much shorter than the typical Intel processor.

With each and every new P4 (Willy, Northwood, Prescott, etc.) Intel continues to lengthen the pipeline as it seeks to create faster and faster processors. Coincidentally, it takes newer P4's greater increases in clock speeds before their features such as increased cache size are notable. It is for this reason that Prescotts generally do not shine until they reach above 3.6-3.7 Ghz.

Think about it as two highways:

The AMD highway is much shorter than that of the Intel. Because the highway is so short, it takes less time for a car to travel the full distance of the highway; hence, there is less of a need for the highway to be very long.

The Intel highway, on the other hand, is much larger than that of the AMD highway. This can be represented by their clock speeds. However, given that the highway is much longer, it takes a car longer to reach the full length of the highway than that of a comparable AMD. It is for this reason why it requires much higher frequencies for an Intel processor to equal that of an AMD (and vice versa).

I appreciate your thought on the matter, but it helps if you have the basics down pat.

Oh, and I'm not a fanboy, either.

deception``

Click to expand...

Continuing with deception's explanation

You can put more cars on the long road that is the P4 than on the Athlon's. However, having more cars (data) on a road increases the chance of a collision. In cpu's however, when a crash (error) occurs, the whole road needs to be cleared. In an AMD processor this takes much less time because there are way less cars than intel. It also takes longer to fill the road up again on intel than on amd for the same reason. The second reason Amd is faster per clock is that it takes less time for AMD's memory controller to tell the stockpile of cars (memory) that it needs more cars for its road.

]-[itman said:

I'll round up some sites for you and pick out the ones to start with, it might take me a day or two(kinda been real busy lately), but I'll pm them to you when I get them.

Click to expand...

Appreciated. thanks

This has got to be the best explanations I have ever read.

No Mumbo Jumbo!

No Acronyms that takes 3 hours of dictionary.com searching to figure out what it means.

Straight forward explanations!!!!!

Consider this the first and probably only OCF thread I bookmark.

Awesome!

Simply Awesome!

Sworkhard said:

Continuing with deception's explanation

You can put more cars on the long road that is the P4 than on the Athlon's. However, having more cars (data) on a road increases the chance of a collision. In cpu's however, when a crash (error) occurs, the whole road needs to be cleared. In an AMD processor this takes much less time because there are way less cars than intel. It also takes longer to fill the road up again on intel than on amd for the same reason. The second reason Amd is faster per clock is that it takes less time for AMD's memory controller to tell the stockpile of cars (memory) that it needs more cars for its road.

Click to expand...

Also why HT works on P4's, but would give marginal improvement on A64's(and the new P-M's). HT uses the dead space kinda "between" the cars of one instruction to start the cars from another, but there isn't very much of that space on the A64 and P-M highways. It all comes down to pipelining which would require a lot of explanation to fully understand it, but this is a good representation of the basic idea of it.

Finally explained in a way I can understand. Thanks!